From: Martin Kronbichler Date: Sat, 24 Aug 2013 02:02:52 +0000 (+0000) Subject: Update comment about no normal flux boundary conditions around reentrant corners X-Git-Tag: v8.1.0~969 X-Git-Url: https://gitweb.dealii.org/cgi-bin/gitweb.cgi?a=commitdiff_plain;h=e25b366a91e558a997df5c40c660f8349626b6ce;p=dealii.git Update comment about no normal flux boundary conditions around reentrant corners git-svn-id: https://svn.dealii.org/trunk@30471 0785d39b-7218-0410-832d-ea1e28bc413d --- diff --git a/deal.II/include/deal.II/numerics/vector_tools.h b/deal.II/include/deal.II/numerics/vector_tools.h index f83d28fc42..5cb317c497 100644 --- a/deal.II/include/deal.II/numerics/vector_tools.h +++ b/deal.II/include/deal.II/numerics/vector_tools.h @@ -1529,7 +1529,13 @@ namespace VectorTools * different boundary indicators and call this function twice, once for each * boundary indicators; doing so will yield only one normal vector at this * point per invocation (because we consider only one boundary part at a - * time), with the result that the normal vectors will not be averaged. + * time), with the result that the normal vectors will not be averaged. This + * situation also needs to be taken into account when using this function + * around reentrant corners on Cartesian meshes. If no-normal-flux boundary + * conditions are to be enforced on non-Cartesian meshes around reentrant + * corners, one may even get cycles in the constraints as one will in + * general constrain different components from the two sides. In that case, + * set a no-slip constraint on the reentrant vertex first. * * *

Computing constraints in 3d